AI and Its Alternatives for Shared Autonomy in Assistive and Collaborative Robotics

Organizers: Aleksandra Kalinowska, Alexander Broad, Brenna Argall, Todd Murphey, Adam Zoss


Shared autonomy is a critical component of human-robot interaction (HRI) that allows robots to collaborate with and provide intuitive assistance to human partners. It is an interdisciplinary domain that brings together the science of human behaviors with state-of-the-art methods in engineering, such as artificial intelligence (AI). Unfortunately, there is often a disconnect between researchers that explore the human side of these interactions and those that propose new engineering solutions. In the former, researchers study human psychology, neuroscience, and biomechanics to better understand how the human partner thinks and functions. In the latter, researchers develop autonomous technologies to improve the capabilities of the human-in-the-loop, often ignoring important facets like perceived utility and acceptance. In this workshop, we bring together experts from both perspectives to define and address challenges in designing and implementing shared autonomy solutions. We will ask questions like — is it possible to create a reliable model of human intent? Can we design interpretable AI to better facilitate HRI? How do we incorporate formal notions of safety with data-driven methods? And finally, how do we take learnings from psychology, neuroscience, and self-driving cars to create better assistive technologies? This workshop will foster multidisciplinary discussion and friendly debate as well as consolidate perspectives, methodologies, and assessment tools to grow research efforts in human-centered robotics.